Recursive prediction error identification of fractional order models

This paper deals with time domain identification of fractional order systems. A new identification technique is developed providing recursive parameters estimation of fractional order models. The identification model is defined by a generalized ARX structure obtained by discretization of a continuous fractional order differential equation. The parameters are then estimated using the recursive least squares and the recursive instrumental variable algorithms extended to fractional order cases. Finally, the quality of the proposed technique is illustrated and compared through the identification of simulated fractional order systems.